View
212
Download
0
Embed Size (px)
Citation preview
The cognitive effects of a mathematics in-service workshop on elementary school teachersAuthor(s): WILLIAM M. BART and ROBERT E. ORTONSource: Instructional Science, Vol. 20, No. 4 (1991), pp. 267-288Published by: SpringerStable URL: http://www.jstor.org/stable/23369963 .
Accessed: 28/06/2014 18:32
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].
.
Springer is collaborating with JSTOR to digitize, preserve and extend access to Instructional Science.
http://www.jstor.org
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
Instructional Science 20: 267-288 (1991) 267 © Kluwer Academic Publishers, Dordrecht - Printed in the Netherlands
The cognitive effects of a mathematics in-service workshop on
elementary school teachers
WILLIAM M. BART & ROBERT E. ORTON
College of Education, University of Minnesota, Minneapolis, MN 54455, U.S.A.
Abstract. Siegler's rule assessment methodology was used to investigate the cognitive effects of a 32
hour mathematics in-service workshop on 28 elementary school teachers. This paper reports on
attempts (a) to assess the levels of cognitive understanding of the basic concepts of proportion, proba
bility, and correlation among elementary teachers, and (b) to change teachers' levels in understanding these concepts by a method of "direct instruction". Significant improvements were noted in the levels
of cognitive development associated with the concept of proportion. Though nonsignificant improve ments were noted in the concepts of probability and correlation, all the teachers were assigned to the
highest cognitive rule associated with the probability concept at the end of the workshop.
Introduction
Mathematics instruction is a major component in elementary education. If ele
mentary school teachers are to be successful in teaching mathematics, then they
need to understand the basic mathematical content that they will be teaching.
Unfortunately, many elementary teachers are deficient in the mathematical
knowledge that they teach their students (Lacampagne et al., 1988; Post et al.,
1988). From a sample of over 200 elementary teachers in a large Midwestern city,
a little over 50% could divide 1/3 by 3, and less than 50% could divide 3 by 4/3
(Post et al., 1988). A remediation program to retrain less knowledgeable teachers in these "basic
skills" might be one solution to this problem. However, mathematics educators
argue that remediation programs that merely reteach the basic skills do not
address the underlying problem, which is a faulty understanding of the concepts and principles behind these basic skills (NCSM, 1978; NCTM, 1980). Ideally, a
remediation program would aim at cognitive training in the concepts underlying
elementary school mathematics.
One set of fundamental concepts underlying elementary school mathematics
are those of proportion, probability and correlation that are identified in the work
of Piaget (Inhelder and Piaget, 1958; Siegler, 1981). Although the concepts are
fundamental, they are not simple one-variable concepts such as height and
weight. These concepts entail the coordination of several variables and thus are
sophisticated concepts.
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
268
In the case of proportion, the concept is applicable to any situation involving
variables a, b, c, and d if the variables satisfy an equality of two ratios such as
a/b = c/d. An example of such a situation involves a balance scale with two
weights in which (a) the weight on the left side of the fulcrum weighs W, ounces
and is Dj inches from the fulcrum, (b) the weight on the right side of the fulcrum
weighs W2 ounces and is D2 inches from the fulcrum, and (c) the scale is level. If
these variables Wj, Dj, W2, and D2 satisfy the proportional relationship
Wj/W2= D2/Dj, then the concept of proportion is applicable to this situation. A
proportional relationship allows one to make correct predictions regarding the
value of one of the variables if one knows the values of the other three variables.
Children with inadequate understanding of these concepts will often ignore one
of the relevant variables and make judgments based on a proper subset of the rele
vant variables. Thus, such a child will say that a balance scale is level if a 5 ounce
weight is placed 2 inches from the fulcrum on the left side of the scale and a
5 ounce weight is placed 4 inches from the fulcrum on the right side, because the
two weights are equal. To such an individual, the distance of either weight from
the fulcrum is irrelevant.
As children develop more sophisticated understanding of multiple variable
concepts such as proportion and probability, they learn to encode all of the rele
vant variables and they learn how the variables are to be coordinated. Inhelder
and Piaget contended that adolescents and adults will have well-developed con
cepts of proportionality, probability and correlation only after they can coordinate
several variables into an integrated system.
Inhelder and Piaget used clinical interviews to assess levels of understanding
of adolescents regarding these three fundamental concepts. Though the interviews
provide some corroboration for their theory, that theory lacks precision in terms
of differentiating among levels of understanding of the concepts. That precision can be garnered by using a detailed information processing model of cognitive
processing. A clear analysis of an intelligence task was provided by Siegler (1976) regard
ing the Inhelder-Piaget balance task. He used information processing methods to
examine the concept of proportionality. Siegler posited that there will be develop mental differences with respect to the balance task, because children at different
ages will demonstrate different levels of understanding of the concept by employ
ing different cognitive rules in their responses to the task. For his task items,
Siegler posited four cognitive rules, each of which indicates a different level of
understanding of the proportionality concept. These cognitive rules are explained in greater detail later in this article.
Siegler's rule-based assessment method, though still debatable, makes possible
precise testing of Inhelder and Piaget's ideas regarding the growth of
sophisticated logical ideas in children, adolescents and adults. This attribute of
"testability" is one of the hallmarks of scientific theorizing (Popper, 1963). The
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
269
results described in this article also indicate that Siegler's work is useful in
assessing the cognitive effects of direct instruction with adult learners. This
attribute of "fruitfulness" (or explanatory power) is another hallmark of scientific
theorizing (Lakatos, 1970). This paper reports on attempts (a) to assess the levels of cognitive understand
ing of the basic concepts of proportion, probability and correlation among elementary teachers, and (b) to change teachers' levels in understanding these
concepts by a method of "direct instruction".
Siegler' s rule assessment methodology
Robert Siegler (1981) fashioned two sets of 24 tasks each of which assessed
understanding of the concepts of proportion and probability, respectively. Each task provided three choices. For example, in any of the probability tasks, the sub
ject would be shown two sets of marbles. In each set (set A or set B), there would
be some white marbles and some black marbles. The subject would be asked
which set of marbles provides the best chances of picking a black marble with the
three choices being (a) set A, (b) set B, and (c) both sets provide the same chance.
By examining the response patterns to the tasks, Siegler was able to classify sub
jects according to four well-defined cognitive rules (or problem solving
strategies) which reflected levels of cognitive development regarding those con
cepts. The rules were termed rules 1, 2, 3, and 4, with rule 1 indicating the lowest level of cognitive development and rule 4 indicating the highest level of cognitive
development.
For each of the three mathematical concepts, the four well-defined cognitive
rules were similar in terms of their relative complexities; they were different in
terms of the component stimulus features being addressed and in terms of the
methods of combining those features. Each of the concepts assessed entailed two
variables, a dominant variable and a subordinate variable. For example, the
proportionality concept assessed by Siegler's balance task items involved the
dominant variable of weight (i.e., the weight on one side of the balance) and the
subordinate variable of distance (i.e., the distance from the fulcrum of the weight on one side of the balance). For each of the four rules associated with any of the
concepts, the values of either the dominant variable or both variables are consid
ered. Rule 1 responses indicated that the subjects attended only to the dominant
variable of the problem situation, whereas rule 4 responses indicated that the sub
jects attended to both dominant and subordinate variables at the same time.
One way to explain these rules is to use flowcharts. For example, as shown in
Figure la, according to rule 1, the subject chooses the side (or set) which has the
greater value for the dominant variable (DL or DR); if the two sides (or sets) have
equal values for the dominant variable, then the subject selects the third choice,
i.e., the choice indicating equivalence between the two sides (or sets).
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
270
According to rule 2 (depicted in Figure lb), the subject chooses the side with
the greater subordinate variable value (SL or SR) only if the dominant variable
values are equal; otherwise, the subject follows rule 1. According to rule 3
(Figure lc), the subject chooses the side (or set) with the greater dominant varia ble value if that side (or set) also has the greater subordinate value or guesses if
the side (or set) with the greater dominant variable value is the side (or set) with the smaller subordinate variable value; otherwise, the subject follows rule 2.
According to rule 4 (Figure Id), the subject employs an algorithm that generates the correct answer if the side (or set) with the greater dominant variable value has
the smaller subordinate variable value; otherwise, the subject follows rule 3.
Awang (1984) developed paper-and-pencil versions of the Siegler task sets and
fashioned a comparable test for the concept of correlation composed of 24 items.
Awang used the tests to examine the levels of understanding of the concepts of
probability, proportionality and correlation among American and Malaysian
college students. Examples of the Awang items are provided in Figures 2-4. The
tasks (test items) associated with the concepts of proportionality, probability and
correlation will be referred to as the balance beam, marbles, and fish tasks,
respectively.
Equal Side with Greater
Dominant Variable
Figure la. Model of rule 1
Does . DL = DR?,
No
SL = SR?.
Side with Greater
Dominant Variable
Yes /
Equal
No
Side with Greater
Subordinate Variable
Figure lb. Model of rule 2
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
271
Greater Dominant Variable
Greater Subordinate
Variable
Side with Greater Guess Dominant and Greater Subordinate Variables
Figure le. Model of rule 3
Does
sDL = m.y
Yes . No
Does SL = SR?.
Does
. SL = SR?,
Yep
Equal
No Yes,
Side with Greater
Subordinate Variable
Side with Greater
Dominant Variable
No
Greater D Same Side ar .Greater S?>
Yes No
Side with Greater Use an Algorithm Dominant and Greater to Determine Side Subordinate Variables
Figure Id. Model of rule 4
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
272
THE BALANCE BEAM PROBLEMS
In the problems to follow there are 24 balance scales with each having equally spaced pegs along their lengths. Each balance scale would have between one and
six metal disks on one of the pegs on each side of the fulcrum. The scale is kept motionless by placing two supports under its arms.
For each problem, please answer the following question:
Which side of the scale would go down, or would it remain level (balance) when supports A and B are removed?
Please mark your answer in the box provided.
For example:
linn linn
* JX i IVVVJ ESS VM kVm
A B
In the above example, right side would go down. The answer would be:
□ left-side down
H right-side down □ remain level (balance)
Please turn the page and proceed carefully. Good luck!
Figure 2. Sample item for the balance task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
273
THE MARBLES PROBLEMS
In each of the problems to follow you will be shown two piles of marbles, some
black ones and some white ones in each pile.
For each problem, please answer the following question:
Which pile of marbles would you want to choose from if you want a black marble and had to choose with your eyes closed?
Please mark your answer in the space provided.
For example:
6&£> c§J8i Pile A Eik B
In the above example, pile B would give us a better chance of picking a black marble than pile A The answer would be:
□ choose pile A IS choose pile B □ choose either one
Please turn the page and proceed carefully. Good luck!
Figure 3. Sample item for the marbles task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
274
THE FISH PROBLEMS
In each of the problems to follow you will be shown two pictures of fish, picture A and picture B. Some of the fish in each picture are big and some are small. Also, some of the fish are black and others are white.
For each problem, please study the two pictures carefully and answer the
following question:
Which picture would you think has a higher relationship or correlation between the size of the fish and the color of their bodies?
Please note that a higher relationship exists if most of the big fish are black and most of the small fish are white, or, if most of the big fish are white and most of the small fish are black.
Please mark your answer in the space provided. For example:
Picture A Picture B
1X3 DO DO
In the above example, picture B shows a higher relationship between the size of
the fish and the color of their bodies than picture A. The answer would be:
□ picture A ®
picture B
□ same in both pictures
Please turn the page and proceed carefully, Good luck!
Figure 2. Sample item for the fish task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
275
The rule sets associated with the three concepts differ somewhat in certain
details from each other, but essentially they all are special cases of the rules
described in Figures la-d. A difference worth mentioning is that one branch of the rule 3 strategy for the proportionality problems involves a guessing
component, whereas the same branch of the rule 3 strategy for the probability and correlation problems involves an additive algorithm. There is evidence that
additive strategies often dominate multiplicative strategies in subjects whose cog nitive schemes are less developed (Karplus, 1981; Karplus, Pulos and Stage,
1983). The rule 4 strategies for the 3 concepts all employ multiplicative algorithms.
The problem solving strategies
Figures 5-7 depict the various rules for the concepts. Figures 5a-d provide the
models for rules 1-4 for the concept of proportionality as assessed by the balance
task. In these models, the following abbreviations are used: (a) LW = the weight on the left side of the balance; (b) LD = the distance from the fulcrum of the
weight on the left side of the balance; (c) RW = the weight on the right side of the
balance; (d) RD = the distance from the fulcrum of the weight on the right side of
the balance.
' Does \
LW = RW?
Yes
Balance
No
Side with Greater
Weight Down
Figure 5a. Model of rule 1 for the concept of proportionality as assessed by the
balance task
Balance Side with Greater Distance Down
Figure 5b. Model of rule 2 for the concept of
proportionality as assessed by the balance task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
276
Side with Greater Guess
Weight and Distance Down
Figure 5c. Model of rule 3 for the concept of proportionality as assessed by the balance task
Balance Side with Greater
Weight x Distance Product
Figure 5d. Model of rule 4 for the concept of proportionality as assessed by the balance task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
277
Figures 6a-d present the models for rules 1^1 for the concept of probability as
assessed by the marbles task. In these models, the following conventions are
used: (a) LB = number of black marbles on the left side; (b) RB = number of
black marbles on the right side; (c) LW = number of white marbles on the left
side; and (d) RW = number of white marbles on the right side.
Equal Side with
Chances More Black
Marbles
Figure 6a. Model of rule 1 for the concept of probability as assessed by the marbles
task
Equal Side with Chances Less
White Marbles
Figure 6b. Model of rule 2 for the concept of
probability as assessed by the marbles task
Equal Side with Chances Greater
B-W Value
Figure 6c. Model of rule 3 for the concept of probability as assessed by the marbles task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
278
Side with Less
White Marbles
Side with More Black
Marbles
Equal Chances
Equal Chances
Side with Greater
B/(B+W) Value
Figure 6d. Model of rule 4 for the concept of probability as assessed by the marbles task
Figures 7a - d provide the models for rules 1-4 for the concept of correlation as assessed by the fish task. In these models, the following conventions are used:
(a) LC = sum of the number of large black fish and the number of small white fish on the left side; (b) LD = sum of the number of large white fish and the num
ber of small black fish on the left side; (c) RC = sum of the number of large black fish and the number of small white fish on the right side; and (d) RD = sum of the number of large white fish and the number of small black fish on the right side.
Does LC = RC?
Yes^
Equal
No
Side with Grealer C
Value
Figure 7a. Model of rule 1 for the concept of correlation as assessed by the fish task
Side with Greater C
Value
Equal Side with Smaller D Value
Figure 7b. Model of iule 2 for the concept of
correlation as assessed by the fish task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
279
Equal Side with Greater C-D Value
Figure 7c. Model of rule 3 for the concept of correlation as assessed by the fish task
Equal Side with Greater
l(C - D)/(C+D)I Value
Figure 7d. Model of rule 4 for the concept of correlation as assessed by the fish task
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
280
Transition to higher rules by direct instruction
Inhelder and Piaget (1958) conjectured that one must pass through a phase of cog nitive disequilibrium prior to attainment of the period of formal operations. In this
phase of cognitive disequilibrium, individuals recognize the inadequacies of prob lem solving strategies that are appropriate to concrete operational thought and
seek more powerful problem solving strategies that are appropriate to formal
operational thought. Rule 4 for each concept assessed in this study entails formal
operational reasoning. Siegler found that cognitive transitions between rule 1 and
rule 2 and between rule 2 and rule 3 for the concept of proportionality, for exam
ple, could be promoted by providing corrective feedback with the test items and/ or by training subjects to encode the subordinate variable.
This study attempted to establish cognitive disequilibrium to promote transition
to rule 4 for each concept. Whereas in Siegler's research, the subjects were chil
dren and adolescents, in this study they were adult teachers. The method used was
"direct instruction", an expository approach wherein teachers were provided an
explanation for inadequacies in their cognitive rules and were given corrective
feedback information (rules and procedures) for rule 4 solution of the test items.
This expository approach also focused on worked or "best" examples of the pro
portion, probability correlation problems. The exact sequence of instruction with
examples is described in more detail below.
Some evidence exists that the direct teaching of a best example or "exemplar model" of a concept can promote its acquisition (Tennyson et al., 1983; Rosch,
1975, 1978; Smith and Medin, 1981). If a best example of a concept is taught in
conjunction with strategy information for the rules and procedures for concept
identification, then concept learning is enhanced even more (Tennyson et al.,
1975). For example, Petkovich (1986) showed that direct teaching of worked
algebra problem examples, together with strategy information, can promote the
acquisition of underlying concepts. This study explores the feasibility of using direct instruction as a means of
facilitating transition through a phase of cognitive disequilibrium to the period of formal operational thought. The general question is: Can direct instruction be used as a method to effect transition in teachers from lower to higher cognitive rules?
Method
Subjects
Twenty eight teachers of grades four, five and six enrolled in a four week
(32 hour) mathematics education workshop on problem solving. Seven teachers were male, and 21 were female. All teachers were currently teaching in the sub urbs of a large Minnesota city.
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
281
Procedure
The teachers were administered three cognitive pretests at the start of the work
shop and three cognitive posttests at the end of the workshop. Midway through
the workshop, or during the fourth class meeting, there was a four hour session on
Siegler's rule assessment approach and the cognitive rules used to solve the
Awang cognitive test items.
Workshop instruction
The session on Siegler's rule assessment approach and the cognitive rules to solve the Awang cognitive test items was placed in the context of teaching teachers about cognitive development. Teachers were told that the Siegler and Awang
work could help them understand how children might think about fundamental
mathematical concepts and also help them, as teachers, think through these ideas. An expository or direct instruction method, combined with strategy informa
tion and several worked examples, was used during the four hour session. Teachers were first given a copy of the balance beam pretest and asked to mark
their answers. After completing the test, the variables of weight and distance were
identified as fundamental variables for the balance beam task. A cognitive rule for the balance beam task was defined as a strategy for deciding whether the bal ance beam would tip to the left, right, or balance, based on the values of the
weight and distance variables.
Teachers were next given a worksheet on which the left (L) and right (R) val
ues for the variables weight (W) and distance (D) were recorded for each item.
Figure 8 shows a copy of this worksheet. Item #1 has a left weight (LW) equal 2, left distance (LD) equal 4, right weight (RW) equal 4, and right distance (RD)
equal 4. Teachers were then told that an examinee who follows rule 1 only attends to the dominant variable of weight. They were then asked to mark, on the rule I column of their worksheet, how an examinee following a rule 1 strategy
would respond for each item.
The procedure for rule 1 was then repeated for rules 2-4. Rule 2 was explained
as a strategy wherein the examinee attends to the subordinate variable, distance,
only if the values for the dominant variable, weight, are equal; otherwise, the
examinee follows rule 1. Rule 3 was explained as a strategy that expands on
rule 2 in which the examinee attends to dominant and subordinate variables for all
cases. In particular, rule 3 helps settle the case where the values of the dominant
variable are not equal but the values of the subordinate variable may or may not
be equal. Teachers were given a copy of the flowchart for rule 3 (similar to
Figure 5c) to help them think through this rule. Rule 4 was explained as a strategy that expands upon rule 3 by replacing the "guessing" or "muddling through" with
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
282
an algorithm wherein distance is multiplied by weight to determine the side with the greatest torque.
For each of the rules 1-4, teachers were asked to respond in the corresponding column how an examinee following that rule would respond for each item (see Figure 8). Thus, teachers had five responses for each item: the values for thinking "as if' they were following rule 1 - rule 4 and their original response for the item before instruction. Based on these data, teachers were then asked to determine which cognitive rule they were most likely following. This activity generated some dissonance and disequilibrium as teachers discussed among themselves which rules they were following and also thought about the possibility of using more sophisticated rules.
The procedure mentioned heretofore of taking the test, identifying dominant
and subordinate variables, determining rule 1 - rule 4 responses, and working through the items "as if' an examinee were following each of the rules, was
repeated for the probability and the correlation pretests.
Expected Answers: Balance Beam Problems
1
2
3
4
5
6
7
8
9
10
11
12
Values Rules Your Answer
LW LD RW RD I n m IV
2 4 4 4
4 4 6 2
3 113
3 2 3 4
4 2 4 2
4 12 4
4 4 6 4
3 4 4 3
4 4 4 3
3 4 14
3 4 5 2
5 2 5 4
Figure 8. Sample worksheet used by teachers during the Direct Instruction session
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
283
Pre- and posttests
The pre- and posttests were constructed by subdividing each of the three Awang tests into two subtests. The odd items in an original Awang test constituted one
subtest of 12 items and the even items constituted another subtest of 12 items.
There were thus six resulting tests: three pretests and three posttests.
The response patterns for each of the six tests were analyzed in terms of the
extent to which they comply with the response patterns produced by each of the
four cognitive rules determined for each concept. Teachers were classified as
complying with a given cognitive rule for a specific test if their response patterns most closely fitted the response pattern generated by that cognitive rule when
compared to the response pattern generated by the other three cognitive rules for
the test.
The classification scheme can be best explained by thinking of a teacher's
score on each of the 4 rules as a vector in 4-space. The four components of this
vector are the teacher's scores on each of the four rules. The score of a teacher
who followed a "pure rule 1" might also be thought of as a vector in 4-space. The
components of the pure rule 1 vector are the scores of a teacher on each of the
four rules, under the assumption that he/she followed a pure rule 1 strategy.
Similarly, there are pure rule 2, pure rule 3, and pure rule 4 vectors. A teacher
was classified by computing the Euclidean distance between his/her score and
each of these four pure rule vectors and selecting that rule whose pure rule vector
was the closest to the teacher's score vector. This might be compared with
Pearson's "least squares criterion", which is used to find the regression line which
best fits a set of data points (cf., Minium, 1978, p. 179).
Unfortunately, the distances between successive pure rule vectors are not
equal. This implies that the rule assignments do not constitute an interval level of
measurement. When testing hypotheses about differences between rule
assignments due to the direct instruction treatment, it was thus necessary to use
nonparametric methods (cf., Siegel, 1956).
Results
One major finding was that not all the elementary school teachers manifested the
highest level of cognitive development for each of the three concepts. There was
substantial variation regarding the cognitive rules being manifested by the teach
ers for each of the three pretests. To be specific, for the balance beam proportion
pretest, 2 teachers were classified as using rule 2, 9 as using rule 3, and 17 as
using rule 4. On the marbles probability pretest, 2 teachers were classified as
using rule 3, and 26 as using rule 4. On the correlation pretest involving "fish"
items, 3 teachers were classified as using rule 1, 1 as using rule 2, 12 as using
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
284
rule 3, and 12 as using rule 4. Thus, there were substantial numbers of teachers
who were either using defective strategies in solving the items or were seemingly inconsistent in the strategy they did use or simply guessed.
As a result of the workshop, there were indicated in the posttest data some
significant gains. On both the balance beam proportion posttest and the marbles
probability posttest, all teachers were classified as using rule 4. On the correlation
posttest involving "fish" items, 2 teachers were classified as using rule 1, 1 as
using rule 2, 13 as using rule 3, and 12 as using rule 4. These shifts between pre and posttests are indicated in Tables 1-3.
Due to the fact that the rules do not constitute an interval scale of measure
ment, the non-parametric test, the sign test, was used to assess the hypothesis that
the probability of a posttest rule placement exceeding the pretest rule placement
for any given subject is .50 (cf., Siegel, 1956). The results are found in Table 4.
For the balance task, the sign test engendered 11 positive shifts out of a total of
11 nonzero shifts, which engendered a p-value of .00. Thus, the posttest rules of
the subjects were significantly greater than the pretest rules. For the marbles task, the sign test engendered 2 positive shifts out of a total of 2 nonzero shifts, which
engendered a p-value of .25. This shift was not significant. For the "fish" task, the
sign test engendered 9 positive shifts and 7 negative shifts out a total of 16 non zero shifts, which engendered a p-value of .40. This improvement was not
significant.
Discussion
The data support the contention that the mathematics in-service workshop, with
its four hour session on Siegler's rule assessment research, results in significant
improvements in the levels of cognitive development regarding the cognitive
scheme associated with the concept of proportion. Though the workshop did not
result in significant improvements in the probability concept, all the teachers were
assigned to rule 4 on the probability posttest. The workshop did not result in a sig nificant improvement, on the average, in the levels of cognitive development regarding the cognitive scheme associated with the concept of correlation.
One reason for the inability of over 50% of the teachers to attain rule 4 for the concept of correlation is the basic complexity of the algorithm in rule 4. The rule 4 algorithm for the concept of proportion involves two encoded variables,
weight and distance, and one procedural step, their product. The rule 4 algorithm for the concept of probability involves two encoded variables and two procedural
steps. The rule 4 algorithm for the concept of correlation involves four encoded variables and seven procedural steps. The teachers who were not able to attain
rule 4 for the concept of correlation may not have had the necessary cognitive capabilities to do so - e.g., inadequate chunking strategies and woiking memory capacities to chunk and encode the necessary variables.
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
285
Table 1. Frequencies of subjects manifesting different cognitive rules for the proportionality concept for the pre- and posttests
Pretest rules Posttest rules
1
2
3
4
0
0
0
0
0
0
0
0
0
0
0
0
0
2
9
17
Table 2. Frequencies of subjects manifesting different cognitive rules for the probability concept for
the pre- and posttests
Pretest rules Posttest rules
12 3 4
1 0 0 0 0
2 0 0 0 0
3 0 0 0 2
4 0 0 0 26
Table 3. Frequencies of subjects manifesting different cognitive rules for the correlation concept for
the pre- and posttests
Pretest rales Posttest rules
1 2 3 4
10 0 12
2 0 0 1 0
3 0 0 7 5
4 2 14 5
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
286
Table 4. Sign test of hypotheses that the probability of a posttest rale placement exceeds the pretest rale placement
Posttest/pretest Concept Rule placement Shift Proportionality Probability Correlation
Positive 11 2 9
Negative 0 0 7
Neither 17 26 12
p-value .00 .25 .40
At least three limitations of the above study need to be conceded. The first is
the "ceiling" effect noted for the proportionality and the probability tasks when
used with adult teachers. The tests were originally designed for younger children, and it was to be expected that many adults would likely score very high on the
measures of proportionality, probability and correlations. Nonetheless, even with
the adult population, there are still some significant variation and shifts in the rule
usage as a result of the workshop intervention. More challenging tests of propor
tionality and probability might result in the ability to assess more accurately transition between rule 3 and rule 4. The correlation test, on the other hand,
showed no ceiling effect, perhaps due to the complexity of the rule for this
concept.
A second limitation pertains to the link between the concepts of proportional
ity, probability and correlation and the Awang pencil-and-paper tests. In their
original investigation of proportionality, probability and correlation, Inhelder and
Piaget (1958) used clinical interviews, which allow for greater variation in subject
response. It might be argued that Inhelder and Piaget's notions of proportionality,
probability, and correlation are much richer than those used in this study. Perhaps the Awang tests do not tap these concepts fully. This limitation, which might be
regarded as one affecting the construct validity of the tests, must be considered.
However, even though the Awang tests may not capture the richness of Inhelder
and Piaget's original work, they do provide objective and reliable measures of
formal reasoning conceptual ability. One might regard the Awang tests as provid
ing operational definitions for the concepts of proportionality, probability, and
correlation.
A third limitation pertains to the conjectural status of Piaget and Inhelder's and
Siegler's theories. This study has not included a full description of the evidence,
pro and con, for either of these theories. A complete evaluation of the concepts of
proportionality, probability and correlation would need to say more about this
evidence and competing theories which might explain the same phenomena.
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
287
On the other hand, the present study does permit, in a sense, a more scientific
examination of the original views of Inhelder and Piaget. The original views
regarding proportionality, probability and correlation were proposed against the
backdrop of Piaget's theory of logical operations. In fact, Inhelder and Piaget
(1958) interpreted levels of understanding of those three concepts in terms of the
logical operational structures which characterize the various cognitive stages within Piagetian cognitive theory.
Though very rich, Piaget's theory of logical operations is difficult to test with
out the extensive clinical interviews done by Inhelder and Piaget. However, the
Awang tests, which permit a precise examination of what might be regarded as
operational definitions of proportionality, probability and correlation, permit the
formulation of hypotheses about these concepts which are falsifiable. This
criterion of "falsifiability" is regarded by philosophers of science as one of the
hallmarks of scientific theorizing (Popper, 1963). Furthermore, the Awang tests
have proved to be useful in classifying the cognitive states of the workshop teach
ers with respect to these concepts of proportionality, probability, and correlation.
This paper indicates, to some degree, that the Siegler rule-based assessment
method influenced by the work of Inhelder and Piaget can be of use in assessing the cognitive effects of direct instruction with adult teachers. This method thus
manifests "fruitfulness" (or explanatory power), which is regarded by some phi
losophers of science as a further hallmark of scientific theorizing (Lakatos, 1970). The concepts of proportionality, probability and correlation are likely prerequi
sites for achievement in disciplines such as physics, chemistry, statistics, and
other secondary and post-secondary subject matter areas in mathematics,
engineering and the physical sciences. These pivotal concepts of proportionality,
probability and correlation are central concepts in formal operational reasoning,
viewed by developmental psychologists as constituting the most advanced stage of cognitive development (Inhelder and Piaget, 1958). This stage of formal opera tions marked by the concepts of proportionality, probability and correlation is
attained by only approximately 30% of the adolescents and adults in the U.S.A.
(e.g., McKinnon and Renner, 1971; Schwebel, 1975). Perhaps there are biologi
cally-imposed limits as to what percentage of adults can become formal reasoners
across a variety of advanced concepts regardless of the intervention techniques that are employed - e.g., direct instruction of cognitive rules.
This study indicates that mathematics in-service workshops can be fashioned
in such a way as to result in significant positive cognitive changes among
elementary school teachers. Cognitive developmental research, coupled with
contemporary advances in cognitive assessment, can be used effectively in the
construction of workshop sessions to deepen and enrich the mathematical under
standings of elementary school teachers. This study also exemplifies how the rule
assessment approach, which is presently so important in the area of cognitive
development, can be successfully applied to study cognitive change among teachers resulting from teacher education projects.
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions
288
References
Awang, M. (1984). Rule usage in solving formal operational problems among Malay and Euro American college students: an evaluation of the rule assessment tests. Unpublished doctoral
dissertation. University of Minnesota, Minneapolis. Inhelder, B. and Piaget, J. (1958). Growth of logical thinking from childhood to adolescence. New
York: Basic Books.
Karplus, R. (1981). Education and formal thought: a modest proposal. In I. E. Sigel (Ed.), New direc tions in Piagetian theory and practice. Hillsdale, NJ: Lawrence Erlbaum Associates.
Karplus, R., Pulos, S. and Stage, E. K. (1983). Early adolescents' proportional reasoning on "rate"
problems. Educational Studies in Mathematics, 14, 219-233.
Lacampagne, C., Post, T., Harel, G. and Behr, M. (1988). A model for the development of leadership and the assessment of mathematical and pedagogical knowledge of middle school teachers. PME NA Proceedings of the Tenth Annual Meeting. DeKalb, IL: PME-NA.
Lakatos, I. (1970). Falsification and the methodology of scientific research programs. In I. Lakatos and A. Musgrave (Eds.), Criticism and the growth of knowledge (pp. 91-196). Cambridge: Cambridge University Press.
Minium, E. W. (1978). Statistical reasoning in psychology and education. New York: John Wiley and Sons.
McKinnon, J. and Renner, J. (1971). Are colleges concerned with intellectual development? American Journal of Physics, 39, 1047-1052.
National Council of Supervisors of Mathematics (1978). Position paper on basic skills. Arithmetic
Teacher, 25 (October), 18-22.
National Council of Teachers of Mathematics (1980). An agenda for action: recommendations for school mathematics of the 1980s. Reston, VA: NCTM.
Petkovich, M. D. (1986). Teaching algebra with worked examples: the effects of accompanying text and range of examples on the acquisition and retention of a cognitive skill. Unpublished Doctoral Dissertation (Department of Educational Psychology), University of Minnesota.
Post, T., Harel, G., Behr, M. and Lesh, R. (1988). Intermediate teachers' knowledge of rational num ber concepts. Madison, WI: National Center for Research in Mathematical Sciences Education.
Popper, K. (1963). Conjectures and refutations: the growth of scientific knowledge. New York:
Harper Torchbook.
Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology: General, 104, 192-233.
Rosch, E. (1978). Principles of categorization. In E. Rosch and B. Lloyd (Eds.), Cognition and cate
gorization (pp. 9-31). Hillsdale, NJ: Lawrence Erlbaum Associates.
Schwebel, M. (1975). Formal operations in first-year college students. Journal of Psychology, 91, 133-141.
Siegel, S. (1956). Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill.
Siegler, R. S. (1976). Three aspects of cognitive development. Cognitive Psychology, 8,481-520.
Siegler, R. (1987). Children's thinking. Englewood Cliffs, NJ: Prentice Hall.
Siegler, R. (1981). Developmental sequences within and between concepts. Monographs of the
Society for Research in Child Development, 44(2, Whole No. 198). Smith, E. and Medin, D. (1981). Categories and concepts. Cambridge, MA: Harvard University Press.
Tennyson, R., Steve, M. and Boutwell, R. (1975). Instance sequence and analysis of instance represen ution in concept acquisition. Journal of Educational Psychology, 76, 821-827.
Tennyson, R., Youngers, J. and Suebsonthi, P. (1983). Acquisition of mathematical concepts by chil dren using prototype and skill development presentation forms. Journal of Educational
Psychology, 75, 280-291.
This content downloaded from 91.213.220.109 on Sat, 28 Jun 2014 18:32:18 PMAll use subject to JSTOR Terms and Conditions